Metal material product manual information extraction method and system and electronic equipment

Abstract

The invention provides a metal material product manual information extraction method and system and electronic equipment. The method comprises the steps that a knowledge base storing metal material information is built; obtaining a metal material table template; obtaining a metal material data manual to extract table data; based on the metal material table template, matching from the table data to obtain filling data; based on the metal material table template, matching from the knowledge base to obtain supplementary data; and fusing the filling data and the supplementary data and filling the table template with the fused filling data and supplementary data. According to the method, the metal material table template input by a client can be structured, the knowledge base storing metal material information is externally stored, the target information of the metal material data manual is extracted, and the metal material data manual is extracted and expanded by effectively combining external knowledge.

Description

Technical Field

[0001] This invention relates to the field of metal material retrieval technology, and in particular to a method, system and electronic device for extracting information from metal material product manuals. Background Technology

[0002] In existing technologies, the extraction of product manual information mainly involves sequential processing and directory structure processing of document information. It cannot dynamically extract key information from complex long texts, nor can it further combine and analyze chart information with text semantics. It is difficult to extract and integrate chart information in complex documents. The main reasons are: 1. Long document information cannot be accurately filtered, and key information in tables cannot be accurately extracted; 2. Material data manuals are not effectively integrated with external knowledge. Summary of the Invention

[0003] In order to address the problems existing in the prior art, the present invention aims to provide a method, system and electronic device for extracting information from metal material product manuals, which can improve the accuracy of information extraction.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: A method for extracting information from product manuals of metallic materials, comprising: Build a knowledge base that stores information about metallic materials; Get a template for a metal materials table; Obtain a datasheet of metallic materials to extract tabular data; Based on the metal material table template, the fill data is obtained by matching the table data; Supplementary data is obtained by matching from the knowledge base based on the metal material table template; The populated data and supplementary data are merged and populated into the table template.

[0005] As a further improvement of the present invention, the step of obtaining a metal material data booklet to extract tabular data includes: At least one of image segmentation technology, target detection technology, character recognition technology, and cross-page table recognition and splicing technology is used to extract tabular data from metal material data manuals.

[0006] As a further improvement of the present invention, the step of matching and obtaining fill data from table data based on a metal material table template includes: Based on the title bar of the metal material table template, character matching and semantic matching are used to obtain the fill data from the table data.

[0007] As a further improvement of the present invention, the step of matching and obtaining fill data from table data based on a metal material table template includes: As a further improvement of the present invention, the step of obtaining supplementary data from the knowledge base based on the metal material table template includes: Calculate the confidence level of the merged filler and supplementary data, and then populate the table template with the confidence level calculation results.

[0008] As a further improvement of the present invention, the present invention also includes the following steps: The knowledge base is updated based on the merged populated data and supplementary data.

[0009] As a further improvement of the present invention, the present invention also includes the following steps: Build a template library for storing metal material table templates; The template library is updated based on the merged fill data and supplementary data.

[0010] This invention also provides a system for extracting information from metal material product manuals, characterized in that the method for extracting the aforementioned information from metal material product manuals includes: The knowledge base processing unit is used to build a knowledge base that stores information about metallic materials. The table processing unit is used to obtain metal material table templates; The table recognition and splicing unit is used to obtain metal material data manuals to extract table data; The large model integration processing unit is used to match and obtain fill data from the table data based on the metal material table template, and to match and obtain supplementary data from the knowledge base based on the metal material table template. The table processing unit is also used to populate the table template with fill data and supplementary data.

[0011] As a further improvement of the present invention, the table data includes: supplementary information; the metal material product manual information extraction system further includes: The confidence calculation unit is used to perform confidence calculations.

[0012] The present invention also provides an electronic device, characterized in that the electronic device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the above-described method for extracting information from metal material product manuals.

[0013] The beneficial effects of this invention are as follows: This invention can construct a metal material form template input by the customer and an external knowledge base storing metal material information, extract the target information of the metal material data manual, effectively combine external knowledge to extract and expand the metal material data manual, integrate the information that the user wants to extract, and improve the accuracy of information extraction. Attached Figure Description

[0014] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0015] Figure 1 This is a flowchart of the method for extracting information from metal material product manuals according to the present invention.

[0016] Figure 2 This is a framework diagram of the metal material product manual information extraction system described in this invention. Detailed Implementation

[0017] To make the technical problems solved by the present invention, the technical solutions adopted, and the technical effects achieved clearer, the technical solutions of the embodiments of the present invention will be further described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention provides a method for extracting information from metal material product manuals, such as... Figure 1 As shown, it includes: S1. Build a knowledge base storing information on metallic materials. This process begins by acquiring information on the metallic materials, including attribute categories, unit conversions, testing conditions, and material applications that will be relevant during extraction.

[0019] S2. Obtain the metal material table template. The metal material table template is obtained by the user by selecting or filling in the template library. The template library stores metal material table templates and can be pre-built.

[0020] S3. Obtain a metal material data booklet to extract tabular data. For example, use at least one of image segmentation technology, target detection technology, character recognition technology, or cross-page table recognition and splicing technology to extract tabular data from the metal material data booklet.

[0021] S4. Based on the metal material table template, fill data is obtained by matching from the table data. Specifically, according to the title bar of the metal material table template, character matching and semantic matching are used to obtain fill data from the table data. For example, if the metal material table template contains tensile strength data that needs to be filled, the tensile strength data of the metal material under various conditions is identified from multiple table data through character or semantic recognition. Then, the corresponding data is filled into the metal material table template.

[0022] S5. Based on the metal material table template, supplementary data is obtained by matching from the knowledge base. The knowledge base is queried to see if there are any physical quantities related to or equivalent to the metal material table template. Some physical quantities or units can be converted to each other. For example, the elastic modulus and Young's modulus are the same physical quantities for metal materials. Through this step, different unit conversions can be identified.

[0023] S6. Merge the populated data and supplementary data into the table template. During this process, a confidence score needs to be calculated for the merged populated and supplementary data. The confidence score calculation result is then populated into the table template for user reference. The formula for calculating the confidence score is: Confidence score = OCR confidence score × Table structure confidence score × Large model consistency score Based on this, the accuracy of data filled in by artificial intelligence can be judged by confidence level when it is subsequently filtered by humans. If, after confidence level calculation, the confidence level of the automatically extracted and filled data into the table template is considered to be low, then these data can be manually verified. If the confidence level is high, then verification is not necessary.

[0024] S7. Update the knowledge base based on the populated data.

[0025] S8. Update the template library based on the supplementary data. If the supplementary data contains data that was not in the original metal material table template, the template library can be updated.

[0026] In addition to data, some metal material data sheets also include material properties. For example, a metal material data sheet might mention that for grade DD12, a cold-rolled steel for stamping, its characteristic is "greater ductility than general-purpose grades, suitable for manufacturing deep-drawn and complex-deformed parts." This information can be stored in a knowledge base and used for future queries or selections of similar materials. There may also be updates and supplements to data or similar tables. In this case, data from metal material data sheet A can be added to the knowledge base. The next time information from metal material data sheet B is retrieved, the relevant information can also be extracted from the knowledge base, and the knowledge base is constantly being dynamically expanded.

[0027] Based on the same inventive concept, this invention provides a system for extracting information from metal material product manuals, such as... Figure 2As shown, the method for extracting information from metal material product manuals as described above includes: a knowledge base processing unit, a table processing unit, a table recognition and splicing unit, a large model integration processing unit, and a confidence calculation unit. The knowledge base processing unit is used to build a knowledge base storing metal material information; the table processing unit is used to obtain metal material table templates; the table recognition and splicing unit is used to obtain metal material data manuals to extract table data; the large model integration processing unit is used to match and obtain fill data from the table data based on the metal material table template, and to match and obtain supplementary data from the knowledge base based on the metal material table template. The table processing unit is also used to fill the fill data and supplementary data into the table template; and the confidence calculation unit is used to perform confidence calculations.

[0028] Furthermore, the knowledge base processing unit pre-stores metal material information through knowledge base construction, including the attribute categories, unit conversions, test conditions, and material uses involved in metal material information extraction. The metal material information will interact and update with the large model integration processing unit, which will facilitate the large model integration processing unit to process the information categories and unit conversions of the target metal material table template. The knowledge base processing unit will also input the confidence calculation unit.

[0029] After the user selects a metal material table template from the template library, the table processing unit interacts with the large model integration processing unit. The table processing unit fills in and expands the template based on the metal material table template and the table data extracted from the metal material data manual, and stores any new templates that may be generated into the table template library.

[0030] The table recognition and splicing unit uses image segmentation technology, target detection technology, character recognition technology, etc. to recognize and extract content from tables in metal material data manuals, including cross-page table recognition and splicing technology.

[0031] The large model integration and processing unit is based on a generative large model to construct a specific information processing process. It can integrate table data extracted by the table recognition and splicing unit, metal material table templates provided by the table processing unit, metal material information provided by the knowledge base processing unit, and information such as specific attribute grouping. The table processing unit fills in the metal material table template selected by the user and supplements the data. At the same time, it feeds back the integrated filled data and supplemented data to the knowledge base processing unit and the table processing unit for updating the knowledge base and the template library.

[0032] The confidence calculation unit performs confidence calculations on some extended information in the final generated table. This facilitates reference for the feasibility of automatically filled data when manually combining and expanding data. It mainly combines the related metal material properties in the table knowledge base, the table information collected from the metal material data manual document, and the table data to be extracted from the table template library selected by the user. The three types of information are integrated and the confidence of the automatically integrated and extracted data is calculated. The confidence of the automatically expanded strongly correlated information is also calculated for user reference.

[0033] The invention will now be described in detail with reference to specific implementation processes: Given a Baosteel hot-rolled steel product manual, the user needs to extract the physical properties of different grades from the manual, including yield strength, tensile strength, density, and elongation. First, the metal material data manual is input into a table recognition and splicing unit. This unit automatically identifies and extracts all tables and their information from the manual. Then, the user selects a metal material table template from a template library. This template includes the grade, yield strength, tensile strength, and elongation. Next, the large-scale model integration and processing unit compares the table data identified from the metal material data manual with the information in the selected template. It determines whether the data is suitable for filling the template. During this process, the large-scale model integration and processing unit interacts with the knowledge base processing unit to query for relevant or equivalent physical quantities from the template library, and to perform unit conversions between physical quantities. Furthermore, the confidence calculation unit calculates the confidence level of all information calculations and filling processes in the knowledge base processing unit, table processing unit, and large model integration processing unit, including the confidence level calculation of recommended strongly relevant data. The fused and confidence-calculated filling data and supplementary data are then filled into the table template to form a complete filled table.

[0034] Based on the same inventive concept, this invention also provides a computer device in this embodiment, the computer device including a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, the at least one instruction, at least one program, code set or instruction set being loaded and executed by the processor to implement the above-mentioned method for extracting information from metal material product manuals.

[0035] The processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0036] The memory can be used to store the computer programs or modules. The processor implements various functions of the assistive terminal device based on mirror neuron therapy by running or executing the computer programs or modules stored in the memory and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0037] In this description, references to terms such as "an embodiment," "example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0038] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style of the specification is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0039] The technical principles of the present invention have been described above with reference to specific embodiments. These descriptions are merely for explaining the principles of the invention and should not be construed as limiting the scope of protection of the invention in any way. Based on this explanation, those skilled in the art can readily conceive of other specific embodiments of the invention without inventive effort, and these embodiments will all fall within the scope of protection of the present invention.

Claims

1. A method for extracting information from product manuals of metallic materials, characterized in that, include: Build a knowledge base that stores information about metallic materials; Get a template for a metal materials table; Obtain a datasheet of metallic materials to extract tabular data; Based on the metal material table template, the fill data is obtained by matching the table data; Supplementary data is obtained by matching from the knowledge base based on the metal material table template; The populated data and supplementary data are merged and populated into the table template.

2. The method for extracting information from metal material product manuals according to claim 1, characterized in that, The step of obtaining a metal material data booklet to extract tabular data includes: At least one of image segmentation technology, target detection technology, character recognition technology, and cross-page table recognition and splicing technology is used to extract tabular data from metal material data manuals.

3. The method for extracting information from metal material product manuals according to claim 1, characterized in that, The step of matching and obtaining fill data from table data based on a metal material table template includes: Based on the title bar of the metal material table template, character matching and semantic matching are used to obtain the fill data from the table data.

4. The method for extracting information from metal material product manuals according to claim 1, characterized in that, It also includes the following steps: Calculate the confidence level of the merged filler and supplementary data, and then populate the table template with the confidence level calculation results.

5. The method for extracting information from metal material product manuals according to claim 1, characterized in that, It also includes the following steps: The knowledge base is updated based on the merged populated data and supplementary data.

6. The method for extracting information from metal material product manuals according to claim 1, characterized in that, It also includes the following steps: Build a template library for storing metal material table templates; The template library is updated based on the merged fill data and supplementary data.

7. A system for extracting information from metal material product manuals, characterized in that, The method for extracting information from metal material product manuals as described in any one of claims 1 to 6 includes: The knowledge base processing unit is used to build a knowledge base that stores information about metallic materials. The table processing unit is used to obtain metal material table templates; The table recognition and splicing unit is used to obtain metal material data manuals to extract table data; The large model integration processing unit is used to match and obtain fill data from the table data based on the metal material table template, and to match and obtain supplementary data from the knowledge base based on the metal material table template. The table processing unit is also used to populate the table template with fill data and supplementary data.

8. The metal material product manual information extraction system according to claim 8, characterized in that, Its features are, The table data includes: supplementary information; the metal material product manual information extraction system also includes: The confidence calculation unit is used to perform confidence calculations.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor to implement the metal material product manual information extraction method as described in any one of claims 1 to 6.

Images